ATP-dependent proteases are responsible for a major portion of the degradation of intracellular proteins in eukaryotic and prokaryotic cells. The E. coli Clp protease is representative of a universal family of ATP-dependent proteases that are composed of a proteolytic core (ClpP) and several ATP-dependent regulatory subunits (ClpA-family members). Interactions between the regulatory subunits and the proteolytic subunit may affect the specificity and activity of the Clp proteases in vivo. Our research has focused on the biochemistry of E. coli Clp protease with the aims of (1) defining the specificity of the protease and the mechanism by which it selects targets in vivo, and (2) defining the functions of ATP in the selection and degradation of proteins. We have succeeded in clearly defining two roles for ATP. ATP acts as an allosteric effector for the assembly of the complex between the regulatory component, ClpA, and the proteolytic component, ClpP. Interaction between ClpA and ClpP in turn alters the active site of ClpP, indicated by the ability to cleave intermediate length peptides. Degradation of large proteins, however, requires ATP hydrolysis, indicating that ATP is required for a second step in degradation. The ATPhydrolysis-dependent step probably involves alterations in the structure of large protein substrates (chaperone function) or changes in the interactions between the protein substrates and the enzyme (translocation function). Direct interaction between ClpA and proteins and peptides is indicated by the effects of these substrates on ATPase activity of ClpA in the absence of ClpP. Selectivity of proteolysis by Clp should involve interactions at both the active and the allosteric sites, since degradation of model peptide substrates indicates the active site of ClpP has rather broad specificity. Site-directed mutagenesis of the ATPase site in domain 2 of ClpA indicates that hydrolysis at that site is not required for assembly of active Clp and cleavage of intermediate length peptides. Since mutants altered in domain 2 cannot degrade large proteins, that domain appears to be responsible for the processive steps in degradation by Clp.